4 Documentation to do: (a little bit at a time)
5 - Document running a test version.
6 - Make sure restore options are documented
7 - Document query file format.
8 - Document static linking
10 Testing to do: (painful)
11 - that console command line options work
12 - blocksize recognition code.
15 - Look at ua_prune.c in detail. Why did JobType work at all??????
16 - Figure out how to allow multiple simultaneous file Volumes on
20 - Enable avoid backing up archive device (findlib/find_one.c:128)
21 - Implement FileOptions (see end of this document)
22 - Implement Bacula plugins -- design API
23 - Make hash table for linked files in findlib/find_one.c:161
24 - Make bcopy read through bad tape records.
25 - Need a verbose mode in restore, perhaps to bsr.
26 - Should we dump a SOS when starting a new tape?
27 - bscan without -v is too quiet -- perhaps show jobs.
28 - Add code to reject whole blocks if not wanted on restore.
29 - Implement multiple simultaneous file Volumes on a single device.
30 - Start working on Base jobs.
31 - Make sure the MaxVolFiles is fully implemented in SD
32 - Flush all the daemon messages at the end of every job.
33 - Check if both CatalogFiles and UseCatalog are set to SD.
34 - Check if we can increase Bacula FD priorty in Win2000
35 - Need return status on read_cb() from read_records(). Need multiple
36 records -- one per Job, maybe a JCR or some other structure with
38 - Figure out how to do a bare metal Windows restore
39 - Fix read_record to handle multiple sessions.
40 - Program files (i.e. execute a program to read/write files).
41 Pass read date of last backup, size of file last time.
42 - Put system type returned by FD into catalog.
43 - Possibly add email to Watchdog if drive is unmounted too
44 long and a job is waiting on the drive.
45 - Strip trailing slashes from Include directory names in the FD.
46 - Use read_record.c in SD code.
47 - Why don't we get an error message from Win32 FD when bootstrap
48 file cannot be created for restore command?
49 - Need to specify MaximumConcurrentJobs in the Job resource.
50 - When Marking a file in Restore that is a hard link, also
51 mark the link so that the data will be reloaded.
52 - Restore program that errors in SD due to no tape reports
53 OK incorrectly in output.
54 - After unmount, if restore job started, ask to mount.
55 - Fix db_get_fileset in cats/sql_get.c for multiple records.
56 - Fix catalog filename truncation in sql_get and sql_create. Use
57 only a single filename split routine.
58 - Make Restore report an error if FD or SD term codes are not OK.
59 - Convert all %x substitution variables, which are hard to remember
60 and read to %(variable-name). Idea from TMDA.
61 - Add JobLevel in FD status (but make sure it is defined).
62 - Make Pool resource handle Counter resources.
63 - Remove NextId for SQLite. Optimize.
64 - Fix gethostbyname() to use gethostbyname_r()
65 - Strip trailing / from Include
66 - Move all SQL statements into a single location.
67 - Cleanup db_update_media and db_update_pool
68 - Add UA rc and history files.
69 - put termcap (used by console) in ./configure and
70 allow -with-termcap-dir.
71 - Enhance time and size scanning routines.
72 - Fix Autoprune for Volumes to respect need for full save.
73 - Fix Win32 config file definition name on /install
74 - No READLINE_SRC if found in alternate directory.
75 - Add Client FS/OS id (Linux, Win95/98, ...).
76 - Test a second language e.g. french.
77 - Compare tape to Client files (attributes, or attributes and data)
78 - Restore to a particular time -- e.g. before date, after date.
79 - Make all database Ids 64 bit.
80 - Write an applet for Linux.
81 - Add estimate to Console commands
82 - Find solution to blank filename (i.e. path only) problem.
83 - Implement new daemon communications protocol.
84 - Remove PoolId from Job table, it exists in Media.
85 - Allow console commands to detach or run in background.
86 - Fix status delay on storage daemon during rewind.
87 - Add SD message variables to control operator wait time
88 - Maximum Operator Wait
89 - Minimum Message Interval
90 - Maximum Message Interval
91 - Send Operator message when cannot read tape label.
92 - Verify level=Volume (scan only), level=Data (compare of data to file).
93 Verify level=Catalog, level=InitCatalog
95 - Add keyword search to show command in Console.
96 - Fix Win2000 error with no messages during startup.
97 - Events : tape has more than xxx bytes.
98 - Restrict characters permitted in a Resource name.
99 - Complete code in Bacula Resources -- this will permit
100 reading a new config file at any time.
101 - Handle ctl-c in Console
102 - Implement LabelTemplate (at least first cut).
103 - Implement script driven addition of File daemon to config files.
104 - Think about how to make Bacula work better with File (non-tape) archives.
106 - see setgroup and user for Bacula p4-5 of stunnel.c
107 - Implement new serialize subroutines
108 send(socket, "string", &Vol, "uint32", &i, NULL)
109 - Audit all UA commands to ensure that we always prompt where possible.
110 - If ./btape is called without /dev, assume argument is a Storage resource name.
111 - Put memory utilization in Status output of each daemon
112 if full status requested or if some level of debug on.
113 - Make database type selectable by .conf files i.e. at runtime
114 - gethostbyname failure in bnet_connect() continues
115 generating errors -- should stop.
116 - Add HOST to Volume label.
117 - Set flag for uname -a. Add to Volume label.
118 - Implement throttled work queue.
119 - Check for EOT at ENOSPC or EIO or ENXIO (unix Pc)
120 - Allow multiple Storage specifications (or multiple names on
121 a single Storage specification) in the Job record. Thus a job
122 can be backed up to a number of storage devices.
123 - Implement dump label to UA
124 - Concept of VolumeSet during restore which is a list
125 of Volume names needed.
126 - Restore files modified after date
127 - Restore file modified before date
128 - Emergency restore info:
130 - Backup working directory
132 - Restore -- do nothing but show what would happen
133 - SET LD_RUN_PATH=$HOME/mysql/lib/mysql
134 - Implement Restore FileSet=
135 - Create a protocol.h and protocol.c where all protocol messages
137 - If SD cannot open a drive, make it periodically retry.
138 - Remove duplicate fields from jcr (e.g. jcr.level and jcr.jr.Level, ...).
139 - Timout a job or terminate if link goes down, or reopen link and query.
140 - Find general solution for sscanf size problems (as well
141 as sprintf. Do at run time?
142 - Concept of precious tapes (cannot be reused).
143 - Make bcopy copy with a single tape drive.
144 - Permit changing ownership during restore.
146 - Restore should get Device and Pool information from
147 job record rather than from config.
148 - Autolabel should be specified by DIR instead of SD.
149 - Find out how to get the system tape block limits, e.g.:
150 Apr 22 21:22:10 polymatou kernel: st1: Block limits 1 - 245760 bytes.
151 Apr 22 21:22:10 polymatou kernel: st0: Block limits 2 - 16777214 bytes.
154 - AutoScan (check checksum of tape)
155 - Format command = "format /dev/nst0"
159 - Seek resolution (usually corresponds to buffer size)
160 - EODErrorCode=ENOSPC or code
161 - Partial Read error code
162 - Partial write error code
163 - Nonformatted read error
164 - Nonformatted write error
165 - WriteProtected error
169 - IgnoreCloseErrors=yes
179 - FD sends unsaved file list to Director at end of job (see
181 - Write a Storage daemon that uses pipes and
182 standard Unix programs to write to the tape.
184 - Need something that monitors the JCR queue and
185 times out jobs by asking the deamons where they are.
186 - Enhance Jmsg code to permit buffering and saving to disk.
187 - device driver = "xxxx" for drives.
188 - restart: paranoid: read label fsf to
189 eom read append block, and go
190 super-paranoid: read label, read all files
191 in between, read append block, and go
192 verify: backspace, read append block, and go
193 permissive: same as above but frees drive
194 if tape is not valid.
196 - Ensure that /dev/null works
197 - File daemon should build list of files skipped, and then
198 at end of save retry and report any errors.
199 - Need report class for messages. Perhaps
200 report resource where report=group of messages
201 - enhance scan_attrib and rename scan_jobtype, and
202 fill in code for "since" option
203 - Need to save contents of FileSet to tape?
204 - Director needs a time after which the report status is sent
205 anyway -- or better yet, a retry time for the job.
206 Don't reschedule a job if previous incarnation is still running.
207 - Figure out how to save the catalog (possibly a special FileSet).
208 - Figure out how to restore the catalog.
209 - Some way to automatically backup everything is needed????
210 - Need a structure for pending actions:
212 - termination status (part of buffered msgs?)
213 - Concept of grouping Storage devices and job can use
214 any of a number of devices
216 Read, Write, Clean, Delete
217 - Login to Bacula; Bacula users with different permissions:
218 owner, group, user, quotas
219 - Store info on each file system type (probably in the job header on tape.
220 This could be the output of df; or perhaps some sort of /etc/mtab record.
223 - Design at hierarchial storage for Bacula.
224 - Implement FSM (File System Modules).
225 - Identify unchanged or "system" files and save them to a
226 special tape thus removing them from the standard
227 backup FileSet -- BASE backup.
228 - Turn virutally all sprintfs into snprintfs.
229 - Heartbeat between daemons.
230 - Audit M_ error codes to ensure they are correct and consistent.
231 - Add variable break characters to lex analyzer.
232 Either a bit mask or a string of chars so that
233 the caller can change the break characters.
234 - Make a single T_BREAK to replace T_COMMA, etc.
235 - Ensure that File daemon and Storage daemon can
236 continue a save if the Director goes down (this
237 is NOT currently the case). Must detect socket error,
238 buffer messages for later.
239 - Enhance time/duration input to allow multiple qualifiers e.g. 3d2h
243 Bacula Projects Roadmap
245 last update 27 November 2002
247 Item 1: Multiple simultaneous Jobs. (done)
250 What: Permit multiple simultaneous jobs in Bacula.
252 Why: An enterprise level solution needs to go fast without the
253 need for the system administrator to carefully tweak
254 timing. Based on the benchmarks, during a full
255 backup, NetWorker typically hit 10 times the bandwidth to
256 the tape compared to Bacula--largely. This is probably due to
257 running parallel jobs and multi-threaded filling of buffers
258 and writing them to tape. This should also make things work
259 better when you have a mix of fast and slow machines backing
262 Notes: Bacula was designed to run multiple simultaneous jobs. Thus
263 implementing this is a matter of some small cleanups and
267 Item 2: Make the Storage daemon use intermediate file storage to buffer data.
268 Deferred -- not necessary yet.
270 What: If data is coming into the SD too fast, buffer it to
271 disk if the user has configured this option.
273 Why: This would be nice, especially if it more or less falls out
274 when implementing (1) above. If not, it probably should not
275 be given a high priority because fundamentally the backup time
276 is limited by the tape bandwidth. Even though you may finish a
277 client job quicker by spilling to disk, you still have to
278 eventually get it onto tape. If intermediate disk buffering
279 allows us to improve write bandwidth to tape, it may make
282 Notes: Whether or not this is implemented will depend upon performance
283 testing after item 1 is implemented.
286 Item 3: Write the bscan program -- also write a bcopy program.
289 What: Write a program that reads a Bacula tape and puts all the
290 appropriate data into the catalog. This allows recovery
291 from a tape that is no longer in the database, or it allows
292 re-creation of a database if lost.
294 Why: This is a fundamental robustness and disaster recovery tool
295 which will increase the comfort level of a sysadmin
296 considering adopting Bacula.
298 Notes: A skeleton of this program already exists, but much work
299 needs to be done. Implementing this will also make apparent
300 any deficiencies in the current Bacula tape format.
303 Item 4: Implement Base jobs.
305 What: A base job is sort of like a Full save except that you
306 will want the FileSet to contain only files that are unlikely
307 to change in the future (i.e. a snapshot of most of your
308 system after installing it). After the base job has been run,
309 when you are doing a Full save, you can specify to exclude
310 all files saved by the base job that have not been modified.
312 Why: This is something none of the competition does, as far as we know
313 (except BackupPC, which is a Perl program that saves to disk
314 only). It is big win for the user, it makes Bacula stand out
315 as offering a unique optimization that immediately saves time
318 Notes: Big savings in tape usage. Will require more resources because
319 the e. DIR must send FD a list of files/attribs, and the FD must
320 search the list and compare it for each file to be saved.
323 Item 5: Implement Label templates
325 What: This is a mechanism whereby Bacula can automatically create
326 a tape label for new tapes according to a detailed specification
327 provided by the user.
329 Why: It is a major convenience item for folks who use automated label
332 Notes: Bacula already has a working form of automatic tape label
333 creation, but it is very crude. The design for the complete
334 tape labeling project is already documented in the manual.
337 Item 6: Write a regression script.
340 What: This is an automatic script that runs and tests as many features
341 of Bacula as possible. The output is compared to previous
342 versions of Bacula and any differences are reported.
344 Why: This is an enormous help in preventing introduction of new
345 errors in parts of the program that already work correctly.
347 Notes: This probably should be ranked higher, it's something the typical
348 user doesn't see. Depending on how it's implemented, it may
349 make sense to defer it until the archival tape format and
350 user interface mature.
353 Item 7: GUI for interactive restore
354 Item 8: GUI for interactive backup
356 What: The current interactive restore is implemented with a tty
357 interface. It would be much nicer to be able to "see" the
358 list of files backed up in typical GUI tree format.
359 The same mechanism could also be used for creating
360 ad-hoc backup FileSets (item 8).
362 Why: Ease of use -- especially for the end user.
364 Notes: Rather than implementing in Gtk, we probably should go directly
365 for a Browser implementation, even if doing so meant the
366 capability wouldn't be available until much later. Not only
367 is there the question of Windows sites, most
368 Solaris/HP/IRIX, etc, shops can't currently run Gtk programs
369 without installing lots of stuff admins are very wary about.
370 Real sysadmins will always use the command line anyway, and
371 the user who's doing an interactive restore or backup of his
372 own files will in most cases be on a Windows machine running
376 Item 9: Add SSL to daemon communications.
378 What: This provides for secure communications between the daemons.
380 Why: This would allow doing backup across the Internet without
381 privacy concerns (or with much less concern).
383 Notes: The vast majority of near term potential users will be backing up
384 a single site over a LAN and, correctly or not, they probably
385 won't be concerned with security, at least not enough to go to
386 the trouble to set up keys, etc. to screw things down. We suspect
387 that many users genuinely interested in multi-site backup
388 already run some form of VPN software in their internetwork
389 connections, and are willing to delegate security to that layer.
392 Item 10: Define definitive tape format.
395 What: Define that definitive tape format that will not change
396 for the next millennium.
398 Why: Stability, security.
400 Notes: See notes for item 11 below.
403 Item 11: New daemon communication protocol.
405 What: The current daemon to daemon protocol is basically an ASCII
406 printf() and sending the buffer. On the receiving end, the
407 buffer is sscanf()ed to unpack it. The new scheme would
408 be a binary format that allows quick packing and unpacking
409 of any data type with named fields.
411 Why: Using binary packing would be faster. Named fields will permit
412 error checking to ensure that what is sent is what the
413 receiver really wants.
415 Notes: These are internal improvements in the interest of the
416 long-term stability and evolution of the program. On the one
417 hand, the sooner they're done, the less code we have to rip
418 up when the time comes to install them. On the other hand, they
419 don't bring an immediately perceptible benefit to potential
420 users. Item 10 and possibly item 11 should be deferred until Bacula
421 is well established with a growing user community more or
422 less happy with the feature set. At that time, it will make a
423 good "next generation" upgrade in the interest of data
429 ======================================================
431 It is somewhat like a Full save becomes an incremental since
432 the Base job (or jobs) plus other non-base files.
434 - New BaseFile table that contains:
435 JobId, BaseJobId, FileId (from Base).
436 i.e. for each base file that exists but is not saved because
437 it has not changed, the File daemon sends the JobId, BaseId,
438 and FileId back to the Director who creates the DB entry.
439 - To initiate a Base save, the Director sends the FD
440 the FileId, and full filename for each file in the Base.
441 - When the FD finds a Base file, he requests the Director to
442 send him the full File entry (stat packet plus MD5), or
443 conversely, the FD sends it to the Director and the Director
444 says yes or no. This can be quite rapid if the FileId is kept
445 by the FD for each Base Filename.
446 - It is probably better to have the comparison done by the FD
447 despite the fact that the File entry must be sent across the
449 - An alternative would be to send the FD the whole File entry
450 from the start. The disadvantage is that it requires a lot of
451 space. The advantage is that it requires less communications
453 - The Job record must be updated to indicate that one or more
455 - At end of Job, FD returns:
456 1. Count of base files/bytes not written to tape (i.e. matches)
457 2. Count of base file that were saved i.e. had changed.
458 - No tape record would be written for a Base file that matches, in the
459 same way that no tape record is written for Incremental jobs where
460 the file is not saved because it is unchanged.
461 - On a restore, all the Base file records must explicitly be
462 found from the BaseFile tape. I.e. for each Full save that is marked
463 to have one or more Base Jobs, search the BaseFile for all occurrences
465 - An optimization might be to make the BaseFile have:
471 This would avoid the need to explicitly fetch each File record for
472 the Base job. The Base Job record will be fetched to get the
473 VolSessionId and VolSessionTime.
474 =========================================================
476 ==========================================================
478 For each Incremental job that is run, there may be files that
479 were found but not saved because they were locked (this applies
480 only to Windows). Such a system could send back to the Director
481 a list of Unsaved files.
483 - New UnSavedFiles table that contains:
487 - Then in the next Incremental job, the list of Unsaved Files will be
488 feed to the FD, who will ensure that they are explicitly chosen even
489 if standard date/time check would not have selected them.
490 =============================================================
494 =============================================================
496 Request For Comments For File Backup Options
499 Subject: File Backup Options
502 A few days ago, a Bacula user who is backing up to file volumes and
503 using compression asked if it was possible to suppress compressing
504 all .gz files since it was a waste of CPU time. Although Bacula
505 currently permits using different options (compression, ...) on
506 a directory by directory basis, it cannot do it on a file by
507 file basis, which is clearly what was desired.
509 Proposed Implementation:
510 To solve this problem, I propose the following:
512 - Add a new Director resource type called FileOptions.
514 - The FileOptions resource will have records for all
515 options that can currently be specified on the Include record
516 (in a FileSet). Examples below.
518 - The FileOptions resource will permit an exclude option as well
519 as a number of additional options.
521 - The heart of the FileOptions resource is the ability to
522 supply any number of ApplyTo records which specify POSIX
523 regular expressions. These ApplyTo regular expressions are
524 applied to the fully qualified filename (path and all). If
525 one matches, then the FileOptions will be used.
527 - When an ApplyTo specification matches an included file, the
528 options specified in the FileOptions resource will override
529 the default options specified on the Include record.
531 - Include records will be modified to permit referencing one or
532 more FileOptions resources. The FileOptions will be used
533 in the order listed on the Include record and the first
534 one that matches will be applied.
536 - Options (or specifications) currently supplied on the Include
537 record will be deprecated (i.e. removed in a later version a
538 year or so from now).
540 - The Exclude record will be deprecated as the same functionality
541 can be obtained by using an Exclude = yes in the FileOptions.
544 The following records can appear in the FileOptions resource. An
545 asterisk preceding the name indicates a feature not currently
549 - Compression= (GZIP, ...)
550 - Signature= (MD5, SHA1, ...)
552 - OneFs= (yes/no) - remain on one filesystem
553 - Recurse= (yes/no) - recurse into subdirectories
554 - Sparse= (yes/no) - do sparse file backup
555 - *Exclude= (yes/no) - exclude file from being saved
556 - *Reader= (filename) - external read (backup) program
557 - *Plugin= (filename) - read/write plugin module
560 - verify= (ipnougsamc5) - verify options
563 - replace= (always/ifnewer/ifolder/never) - replace options currently
565 - *Writer= (filename) - external write (restore) program
569 Currently options specifying compression, MD5 signatures, recursion,
570 ... of a FileSet are supplied on the Include record. These will now
571 all be collected into a FileOptions resource, which will be
572 specified on the Include in place of the options. Multiple FileOptions
573 may be specified. Since the FileOptions contain regular expressions
574 that are applied to the full filename, this will give the ability
575 to specify backup options on a file by file basis to whatever level
584 Include = compression=GZIP signature=MD5 {
593 Include = FileOptions=Opts {
604 That's a lot more to do the same thing, but it gives the ability to
605 apply options on a file by file basis. For example, suppose you
606 want to compress all files but not any file with extensions .gz or .Z.
607 You could do so as follows:
611 Include = FileOptions=NoCompress FileOptions=Opts {
619 ApplyTo = /*.?*/ # matches all files
624 # Note multiple ApplyTos are ORed
625 ApplyTo = /*.gz/ # matches .gz files */
626 ApplyTo = /*.Z/ # matches .Z files */
629 Now, since the NoCompress FileOptions is specified first on the
630 Include line, any *.gz or *.Z file will have an MD5 signature computed,
631 but will not be compressed. For all other files, the NoCompress will not
632 match, so the Opts options will be used which will include GZIP
636 - Is it necessary to provide some means of ANDing regular expressions
637 and negation? (not currently planned)
639 e.g. ApplyTo = /*.gz/ && !/big.gz/
641 - I see that Networker has a "null" module which, if specified, does not
642 backup the file, but does make an record of the file in the catalog
643 so that the catalog will reflect an exact picture of the filesystem.
644 The result is that the file can be "seen" when "browsing" the save
645 sets, but it cannot be restored.
647 Is this really useful? Should it be implemented in Bacula?
650 After implementing the above, the user will be able to specify
651 on a file by file basis (using regular expressions) what options are
652 applied for the backup.
653 ====================================
655 Done: (see kernsdone for more)
656 - Add EOM records? No, not at this time. The current system works and
658 - Add VolumeUseDuration and MaximumVolumeJobs to Pool db record and
660 - Add VOLUME_CAT_INFO to the EOS tape record (as
661 well as to the EOD record). -- No, not at this time.
662 - Put MaximumVolumeSize in Director (MaximumVolumeJobs, MaximumVolumeFiles,
664 - Enhance schedule to have 1stSat, ...
665 - Make sure catalog doesn't keep growing.
666 - On I/O error, write EOF, then try to write again ? No, keep it simple.
667 - Figure out how compress everything except .gz,... files.
668 Implement FileOptions.
669 - Put Bacula version somewhere in Job stream, probably Start Session Labels.
670 - Fix start/end blocks for File devices
671 - Make Job err if WriteBootstrap fails.
672 - Test that mod of restore options works.
673 - Test that week position schedule code works.
674 - Make BSR accept count (total files to be restored).
675 - Add code to fast seek to proper place on tape/file when doing Restore.
676 - Replace popen() and pclose() -- fail safe and timeout, no SIG dep.
677 - Add code to put VolFile in bsr for restore command.
678 - Volumes can be listed multiple times in Restore volume list.
679 - Add watchdog timeout for child processes start_child_timer()
681 - Get rid of bscan.c:534 error message (one time only).
682 - Print some statistics when get EOF on device in bscan -- feedback
683 to let user know it is working.
684 - DateWritten field on tape may be wrong.
685 - Ensure that restore of differential jobs works (check SQL).
686 - Count number of ignored messages in bscan and print when first SOS is found.
687 - Test that EndFile/Block are correctly updated at end of tape
688 (in view of new block reading code).
689 - Test watchdog child timer code.
690 - Test new BSR code (mostly done).
691 - Work more on how to to a Bacula restore beginning with
692 just a Bacula tape and a boot floppy (bare metal recovery).
693 - Restore options (overwrite, overwrite if older,
694 overwrite if newer, never overwrite, ...)
695 - Fill all fields in Vol/Job Header -- ensure that everything
696 needed is written to tape. Think about restore to Catalog
697 from tape. Client record needs improving.
698 - Implement ./configure --with-client-only
699 - Finish up static linking
700 - that restore options work in FD